A system and method for recovering from radiation induced memory errors invalidates information stored in a cache memory, upon the detection of the memory error. The cache memory is then reloaded with valid information.
|
1. A method for recovering from a radiation induced memory error, comprising the steps of:
detecting a memory error in a cache memory;
providing an error indication of the memory error;
processing the error indication to determine whether a hardware recovery or a software recovery should be implemented and, based on the determination, providing an associated processor appropriate command to implement either the hardware recovery or the software recovery;
invalidating stored information in response to the associated processor appropriate command; and
reloading the memory with valid information.
7. A processing system that automatically recovers from radiation induced memory errors, the system comprising:
a processor core;
a memory coupled to the processor core, wherein the memory stores information and includes parity circuitry for detecting a memory error in the memory; and
error processing logic coupled to the memory, wherein the error processing logic, upon the detection of the memory error, determines whether a hardware recovery or a software recovery should be implemented and, based on the determination, provides a signal to the processor core which causes the processor core to replace the stored information with valid information using either the hardware recovery or the software recovery.
12. A processing system that automatically recovers from radiation induced memory errors, the system comprising:
a processor core;
a cache memory coupled to the processor core, wherein the memory stores information and includes parity circuitry for detecting a memory error in the memory; and
error processing logic coupled to the memory, wherein the error processing logic, upon the detection of the memory error, determines whether a hardware recovery or a software recovery should be implemented and, based on determination, provides a signal to the processor core which causes the processor core to replace the stored information with valid information using either the hardware recovery or the software recovery.
2. The method of
3. The method of
4. The method of
8. The system of
9. The system of
13. The system of
14. The system of
|
The present invention is generally directed to recovering from memory errors and, more specifically, to recovering from radiation induced memory errors.
Today, commercially available microprocessors are generally not designed to operate in a space borne environment. As such, these commercial microprocessors are subject to radiation induced errors. For example, single event upsets (SEUs) can occur when an ionized particle hits a flip-flop or a memory cell, associated with the microprocessor, and changes the state of the associated flip-flop or memory cell. These radiation induced errors can result in incorrect calculations and faulty program execution, as well as system reset and a loss of state, due to timeout of a watchdog timer. Further, even when a microprocessor is radiation hardened, the microprocessor is still subject to SEUs of internal flip-flops and/or memory cells.
In a memory that includes a parity generator/checker circuit (and when parity checking is enabled), each time a data byte, i.e., eight bits, is written to the memory the circuit examines the byte and determines whether the byte has an even or odd number of ‘ones’. In the case of odd parity, when the data byte has an even number of ‘ones’ a parity bit, i.e., a ninth bit, is set to ‘one’. Otherwise, the parity bit is set to ‘zero’. The result is that no matter how many ‘ones’ were in the original eight bits of data, there are an odd number of ones when all nine bits are examined. Alternatively, instead of implementing odd parity the circuit may implement even parity such that the sum of the ‘ones’ is an even number. In a typical microprocessor system, when a byte is read from memory the circuit checks the parity of the byte to determine whether a parity error is indicated.
In a typical microprocessor system, when a parity error is detected a parity checker/generator circuit generates a non-maskable interrupt (NMI), which is usually used to instruct a microprocessor to immediately halt. This is done to ensure that invalid data does not corrupt valid data. In many microprocessor systems, a watchdog timer, which can be implemented in hardware or software, may be the only means for detecting when an execution error occurs. Alternatively, the watchdog timer may also be implemented in conjunction with a parity generator/checker circuit to detect memory errors. In either case, the microprocessor generally executes code until the watchdog timer times-out or an unrecoverable software occurs.
Microprocessors that have an internal cache memory with a parity generator/checker circuit may provide an output from the circuit off-chip and may also include an external flush line, which causes the microprocessor to flush its internal cache when it receives an appropriate signal. However, in general, space borne microprocessor systems have only used watchdog timers to detect memory errors attributable to SEUs. As a result, for microprocessor systems used in space born environments, the time period between when an application error occurs and the microprocessor system recovers from the error has generally been relatively long and has required resetting the microprocessor system.
What is needed is a technique for recovering from radiation induced memory errors that is both efficient and timely. It would also be desirable to recover from a radiation induced memory error without resetting the microprocessor system.
The present invention is directed to a system and method for recovering from a radiation induced memory error. Initially, a memory error in a cache memory is detected and information currently stored in the cache memory is invalidated. Then, the cache memory is reloaded with valid information.
These and other features, advantages and objects of the present invention will be further understood and appreciated by those skilled in the art by reference to the following specification, claims and appended drawings.
A relatively large percentage (e.g., approximately ninety percent) of single event upsets (SEUs), i.e., radiation induced errors, are attributable to errors in memory cells of an internal (i.e., level 1 (L1) cache memory. While the discussion herein is primarily directed to an L1 cache memory, the techniques disclosed herein are also applicable to external cache memories (e.g., level 2 (L2) cache memories). According to the present invention, when an SEU is detected by an error detection circuit (e.g., a parity checker/generator circuit associated with an L1 cache memory), error processing logic provides a command to a processor core (i.e., an execution unit), which causes the core to flush the L1 cache and reload the L1 cache with valid information.
Turning to
Accordingly, a technique has been described herein that allows a processor to recover from SEU induced memory errors in a relatively efficient, rapid manner. Thus, reducing the likelihood of miscalculations and failed program execution. Such a system and method can also prevent the loss of state due to a watchdog timer timeout, which occurs when the processor system resets. It should be appreciated that the flushing method is processor dependent. That is, while
The above description is considered that of the preferred embodiments only. Modification of the invention will occur to those skilled in the art and to those who make or use the invention. Therefore, it is understood that the embodiments shown in the drawings and described above are merely for illustrative purposes and not intended to limit the scope of the invention, which is defined by the following claims as interpreted according to the principles of patent law, including the Doctrine of Equivalents.
Nelson, William S., DeRuiter, John L.
Patent | Priority | Assignee | Title |
10901097, | Mar 05 2018 | XILINX, Inc. | Method and apparatus for electronics-harmful-radiation (EHR) measurement and monitoring |
12056004, | Sep 02 2021 | Samsung Electronics Co., Ltd. | Method and apparatus with cosmic ray fault protection |
8397130, | Nov 26 2008 | Arizona Board of Regents For and On Behalf Of Arizona State University | Circuits and methods for detection of soft errors in cache memories |
8397133, | Nov 26 2008 | Arizona Board of Regents, for and on Behalf of Arizona State University | Circuits and methods for dual redundant register files with error detection and correction mechanisms |
8489919, | Nov 26 2008 | Arizona Board of Regents, for and on Behalf of Arizona State University | Circuits and methods for processors with multiple redundancy techniques for mitigating radiation errors |
8909981, | Jun 21 2011 | Hamilton Sundstrand Corporation | Control system software execution during fault detection |
9292378, | May 29 2014 | TELEDYNE SCIENTIFIC & IMAGING, LLC | Single event upset protection circuit and method |
RE42314, | Sep 08 2003 | Space Micro, Inc. | Functional interrupt mitigation for fault tolerant computer |
Patent | Priority | Assignee | Title |
5678020, | Jan 04 1994 | Sony Corporation of America | Memory subsystem wherein a single processor chip controls multiple cache memory chips |
5778433, | Mar 22 1993 | HEWLETT-PACKARD DEVELOPMENT COMPANY, L P | Computer system including a first level write-back cache and a second level cache |
6108758, | Aug 29 1997 | Intel Corporation | Multiple masters in a memory control system |
6195729, | Feb 17 1998 | International Business Machines Corporation | Deallocation with cache update protocol (L2 evictions) |
6463507, | Jun 25 1999 | International Business Machines Corporation | Layered local cache with lower level cache updating upper and lower level cache directories |
6751720, | Jun 10 2000 | Hewlett Packard Enterprise Development LP | Method and system for detecting and resolving virtual address synonyms in a two-level cache hierarchy |
20030131277, | |||
20040078702, | |||
20040139280, | |||
EP128353, | |||
EP300166, | |||
EP321793, |
Executed on | Assignor | Assignee | Conveyance | Frame | Reel | Doc |
Mar 26 2002 | DERUITER, JOHN L | Honeywell International Inc | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 012749 | /0457 | |
Mar 26 2002 | NELSON, WILLIAM S | Honeywell International Inc | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 012749 | /0457 | |
Mar 28 2002 | Honeywell International Inc. | (assignment on the face of the patent) | / |
Date | Maintenance Fee Events |
May 12 2005 | ASPN: Payor Number Assigned. |
Sep 18 2008 | M1551: Payment of Maintenance Fee, 4th Year, Large Entity. |
Jan 14 2013 | REM: Maintenance Fee Reminder Mailed. |
May 31 2013 | EXP: Patent Expired for Failure to Pay Maintenance Fees. |
Date | Maintenance Schedule |
May 31 2008 | 4 years fee payment window open |
Dec 01 2008 | 6 months grace period start (w surcharge) |
May 31 2009 | patent expiry (for year 4) |
May 31 2011 | 2 years to revive unintentionally abandoned end. (for year 4) |
May 31 2012 | 8 years fee payment window open |
Dec 01 2012 | 6 months grace period start (w surcharge) |
May 31 2013 | patent expiry (for year 8) |
May 31 2015 | 2 years to revive unintentionally abandoned end. (for year 8) |
May 31 2016 | 12 years fee payment window open |
Dec 01 2016 | 6 months grace period start (w surcharge) |
May 31 2017 | patent expiry (for year 12) |
May 31 2019 | 2 years to revive unintentionally abandoned end. (for year 12) |